Upgrading from TM1 10.2.2 to Planning Analtyics Local ( is PAW required? )

[7Zark7]

Hey Guys,

Appologes for asking this question if it has already been answered. We are looking to upgrade from TM1 10.2.2. to Planning Analtyics Local shortly and wanted to understand if PAW is required for PAX. My understanding is that PAW is an optional and is a seperate install and is not a required to take advantage of PAX functionality?

A few years ago we embarked on this and I believe there is one componeent that we cannnot take advantage of in PAX and that is alternate hierarhies UNLESS you have PAW installed. Is that correct?

If there is existing documentation found in the tips sections / useful IBM documentation and any caveats I should be aware of when upgrading I would be greatly appreciative. Any gotchas that I should be aware of i.e addtional ram consumption required etc from user experience would he helpful as well.

Best

Zark

[Alan Kirk]

Regular viewers may wish to find a comfy chair and put the popcorn in the microwave about now.

Appologes for asking this question if it has already been answered. We are looking to upgrade from TM1 10.2.2. to Planning Analtyics Local shortly and wanted to understand if PAW is required

It is absolutely not required... oh, wait...

for PAX.

Ah. In that case, the answer is yes. And it p*sses me off.

As I understand it the only thing that PAX (I'm not sure whether that's its name this week, but we all know what we're referring to here) is directly dependent on PAW for is for its subset editor. But obviously, a query tool without a subset editor won't be able to do much.

A charitable interpretation of this might be that the team that built PAX just wanted to ensure that users had a standard user experience. A less charitable one might be that they were just too lazy to make PAX completely stand alone. I of course could not possibly say which is the correct interpretation, but I can speculate. And yes, it is entirely possible to make PAX a standalone add in given that there is no need whatsoever to have a PAW implementation to run, for example, Cubewise Arc. "Gee look ma, I can build a complete UI without being dependent on the server side to provide some of it".

My understanding is that PAW is an optional and is a seperate install and is not a required to take advantage of PAX functionality?

Yes, it is optional. Yes, it is a separate install. No, I'm afraid it IS required for PAX as I understand it, unless that has changed recently.

The key problem is the installation of PAW. To run PAW you need to use Docker, a containerisation solution which is kinda sorta like a mini virtual Linux machine. (Yes, I know the difference between those and no pedantry will be entered into; I'm just describing the effect.) IBM actually recommends that you run it on a separate server machine (or did) from the TM1 server. You can in theory run it on Windows but as far as I am aware it is still the case that it is only officially supported on some flavours of Windows Server 2016. No, not 2019. Yes, 2019 is over 2 years old.

The problem is with the installation on Windows which is nothing short of a train wreck. I have never gotten it to run AND CONNECT to anything. Which is to say, yes, I've made it run in the past, no, it wouldn't connect to anything. Though on many occasions I just have not gotten it to run. See also this thread for just how utterly fricking ridiculous the install process can be.

We were at one point told "In IBM Planning Analytics Workspace Local version 2.0.44 or later, it is much simpler to install on Windows Server 2016". My own personal experience post-release 44 was that I just got a new and exciting brace of different PowerShell error messages, each as cryptic and uninformative as their predecessors. But hey, at least they were NEW!

Contrary to popular opinion, my issue is not with Docker. I have run a bunch of RDBMSs of various flavours in Docker containers on Windows 10 for development projects. It's easy, it's simple, it works. My issue is with the piece of **** known as PAW, which I find to be none of the above, and exclusively with that.

A few years ago we embarked on this and I believe there is one componeent that we cannnot take advantage of in PAX and that is alternate hierarhies UNLESS you have PAW installed. Is that correct?

No. You can turn on alternative hierarchies in your server config file regardless of whether PAW is installed. The problem is that without the PAW/PAX combination you won't have any Excel- integrated way to get to them. I could do that now and use Cubewise Arc to create and query them. I could use the API to create and query them, and/or do it through Python, C#, whatever. But I can't use PAX because I can't get PAW running, and alternative hierarchies won't ever be usable in the classic Perspectives and Web clients.

But then, there's also the issue of the absence of hierarchy aware DBRW formulas that IBM haven't been able to bother to provide for... hmm, how many years now? After how many hundreds of requests from users?

"Oh, yeah, we'll get around to it, someday, maybe. Send us an e-mail about it."

For practical purposes, alternative hierarchies won't be of much use to you without PAW and PAX.

If there is existing documentation found in the tips sections / useful IBM documentation and any caveats I should be aware of when upgrading I would be greatly appreciative. Any gotchas that I should be aware of i.e addtional ram consumption required etc from user experience would he helpful as well.

I didn't really check the memory usage thing, but I went from 9.5.2 to PA2 (without PAW/PAX) without any real issues. However if you DO switch on alternative hierarchies you may find that it's not a like for like comparison.

[jim wood]

The client I'm working at recently upgraded from 10.2.2 to PA 2.0.7. (BTW avoid 2.0.7 there's a server side bug that caused random crashes for us so we had to move to 2.0.9) We did not install Workspace at all. We are continuing (for now) to use both TM1 Web and Perspectives. But you do need to keep in mind that both of them will be hitting the sunset in the non too distant future. So then you're left with PAX, which as Alan stated (Sorry Alan I didn't read all your post as it's late and I'm tired) you need Workspace for it to work. We were put off by this as the client doesn't have a Workspace license. That was until we found out that you don't need a Workspace license to install it if you're only using it for PAx. After all certain CA licenses come with PAx licenses built in, so how else would you be able leverage it?

So I hope that helps as well. We're going to look at the improved Windows Workspace installer rather than the Linux version. Another thing to keep in mind,

Jim.

[lotsaram]

...that is until we found out that you don't need a Workspace license to install it if you're only using it for PAx. After all certain CA licenses come with PAx licenses built in, so how else would you be able leverage it?

The other caveat is that Workspace license is only required for end users. As Workspace is the new admin / modelling environment to replace Architect (and Performance Mudler) any users who have an Admin license have automatic rights to use Workspace.

[7Zark7]

Hey Guys,

Alan - thanks for taking the time to clarify my points and the write up of another epic post! I feel your frustration and its dissapointing that the IBM documentation it not exactly clear. Its misleading that PAX requires PAW to be running / installed for us to use key functionality - its dissapointing...

So for the sake of having an environment that will be covered by IBM when the next microsoft update breaks TM1 perspectives ( been lucky so far ), do I just simply upgrade tm1 to Planning Analytics Local and ignore the PAX and PAW components, Just to keep things simple? And then once we have decided whether we would like to shell out coin for this additonal functionality then take the leap? Is this a sensible approach?

Also, thanks lotsaram and Jim for your feedback. Jim, its sad that you embarked on the install only to be bitten at the end in regards to coupling of both PAX and PAW being required to take advantage of key functionality.

Appreciate the input as always.

Best

Zark

[7Zark7]

I assume that perspectives and tm1 web will still work as they always have? Any ideas in regards to when perspectives and tm1 web will be shelved in favour of PAX / PAW combo?

Zark

[Alan Kirk]

So for the sake of having an environment that will be covered by IBM when the next microsoft update breaks TM1 perspectives ( been lucky so far ), do I just simply upgrade tm1 to Planning Analytics Local and ignore the PAX and PAW components, Just to keep things simple? And then once we have decided whether we would like to shell out coin for this additonal functionality then take the leap? Is this a sensible approach?

Yes. Certainly it's what we've done. As per Lotsa's post I'm not sure you'd need to shell out coin if PAW is only used by your existing admin(s); it depends on your licence.

With regard to your follow up, legacy Perspectives (and the classic API) are expected to die in the next major release; i.e. server version 12, which will probably be dubbed PA3.

[Ajay]

I assume that perspectives and tm1 web will still work as they always have? Any ideas in regards to when perspectives and tm1 web will be shelved in favour of PAX / PAW combo?

Zark

So we upgraded last year from TM1 10.2 to PA and we completely ditched Perspectives, and went with PAfE.

PAW - besides providing the connection for PAfE - will improve the performance of your offering to the users

We were in that place where losing Perspectives felt alien.....but I am so glad we did.

For us the benefit was no longer having sh*t performance across the WAN, and having to support a Citrix environment. Docker (through PAW) will take care of this. Perspectives in PA will still have the issue of poor performance. I see that right now myself.

The newer front end has also been a tangible benefit to end users who really like the fresher update.

What's making you keep Perspectives ?

[Alan Kirk]

I assume that perspectives and tm1 web will still work as they always have? Any ideas in regards to when perspectives and tm1 web will be shelved in favour of PAX / PAW combo?

Zark

So we upgraded last year from TM1 10.2 to PA and we completely ditched Perspectives, and went with PAfE.

PAW - besides providing the connection for PAfE - will improve the performance of your offering to the users

We were in that place where losing Perspectives felt alien.....but I am so glad we did.

For us the benefit was no longer having sh*t performance across the WAN, and having to support a Citrix environment. Docker (through PAW) will take care of this.

Let's be clear about one thing. Neither PAW nor Docker have the first thing to do with the improvements in performance that you can get compared to Perspectives. I don't dispute that PAX/PAfE is much better performing than Perspectives, but neither PAW nor Docker have the first thing to do with this.

Containerisation has a number of advantages. When it is done well, an important proviso that some in IBM seem to have missed, it provides easy "lift and shift" capabilities, which is why I use it for RDBMS development. It is also easier to provide 24 hour up time, which is why version 12 of the server will be moving to it. But speed compared to a well written native application is NOT generally one of those advantages. If the native application is done well, and the Docker container is well created, there should not be a huge gain or loss in performance between the two. Or certainly I've yet to find one in SQL Server which is one of the few that I use in both modes.

The reason that PAX is faster is down to the fact that (a) it uses the new API which is inherently more efficient, and (b) it does away with all of the archaic network chattiness that is built into Perspectives.

That is why I can get massive performance improvements compared to Perspectives in Cubewise Arc or doing direct pulls through the API too, without a single instance of PAW or Docker in sight.

The newer front end has also been a tangible benefit to end users who really like the fresher update.

What's making you keep Perspectives ?

The end users will benefit from speed and alternative hierarchies but will also need a fair amount of retraining. That in itself is a good reason to make the transition in two stages.

If you've managed to get PAW to run for you (and as an aside I believe that this is easier on Linux than on Windows, which doesn't help in an all-Windows shop) then more power to you. But as can be seen from a lot of threads including the one that I linked to in my original reply, it's a complete time sink and blood pressure elevation tool for many people in its current state.

My preference would be to run PAX as well, but there aren't enough hours in the day for that cr@p.

[ykud]

If you've managed to get PAW to run for you (and as an aside I believe that this is easier on Linux than on Windows, which doesn't help in an all-Windows shop) then more power to you.

Night and day on Windows and Linux. PaW on Windows is a real box full of unknowns and surprises and doesn't play nicely with Windows updates to containerisation (hence no 2019 support in sight) and has a tendency to die on every anti-virus update. I don't think it's entirely IBM's fault, containers are a bit foreign to Win in general. But you definitely have normal working containers on windows all over the place now (especially with WSL 2) and PaW stands out in this regard.

Linux, on the other hand, is quite straight-forward. Haven't had a problem with it this far. Having said that, only RHEL 7 is supported with built-in docker and if you want to go RHEL 8 (which is exactly like 2016 and 2019 windows), it's not supported yet either

[7Zark7]

Thanks Yuri.

I was not aware of WSL 2 ( until I read up on it just then! ) per : https://docs.docker.com/docker-for-windows/wsl/

So what you are saying is that runing PAW on a windows env with WSL 2 is the way to go and its all peaches and cream? Would you go with this over Linux? If you would go with Linux over Windows with WSL 2, what would be a good reason to do so when setting up PAW?

Thanks

Zark

[Alan Kirk]

I think you may be starting to jump on things without reading them slowly and in context. This could lead you into some perilous plans.

I was not aware of WSL 2 ( until I read up on it just then! ) per : https://docs.docker.com/docker-for-windows/wsl/

So what you are saying is that runing PAW on a windows env with WSL 2 is the way to go and its all peaches and cream?

What he said was that containerisation is becoming quite common with WSL2. What he also said (as I had previously) is:

hence no 2019 support in sight

Looking at the document that you linked to, which version of Windows is it referring to?

Windows 10, not Windows Server of any version.

Prerequisites
Before you install the Docker Desktop WSL 2 backend, you must complete the following steps:

Install Windows 10, version 1903 or higher.
Enable WSL 2 feature on Windows.

(My emphasis.)

In reality, WSL2 is also available in Windows Server 2019. (Or at least certain builds of it.)

BUT... as noted above, PAW itself is not supported on Server 2019, only on certain flavours of 2016. Until or unless that changes, WSL2 is not your magic key.

Would you go with this over Linux? If you would go with Linux over Windows with WSL 2, what would be a good reason to do so when setting up PAW?

From what I have heard (and not just from Yuri, but from a number of people), using Linux if you have that option would reduce the number of times you will need to repaint your office after you have slammed your head into the wall in frustration from trying to get it to run in a relatively archaic Windows environment (Server 2016) which was never really designed to support containerisation but where it was just bolted on as an after-market accessory.

Note that I am not in any way excusing for IBM for PAW being what it is (implementing it in Docker before the main O/S that people use was really ready for it was their decision), but I am prepared to concede that Yuri made an excellent point here:

[PaW on Windows is a real box full of unknowns and surprises and doesn't play nicely with Windows updates to containerisation (hence no 2019 support in sight) and has a tendency to die on every anti-virus update. I don't think it's entirely IBM's fault, containers are a bit foreign to Win in general.

[ykud]

So what you are saying is that runing PAW on a windows env with WSL 2 is the way to go and its all peaches and cream? Would you go with this over Linux? If you would go with Linux over Windows with WSL 2, what would be a good reason to do so when setting up PAW?

Alan is spot on, I didn't say anything about peaches or cream, quite far from it. Windows 2019 (with updates to containers making them passable) is not supported by PaW as of now.

Installing PaW on Windows is an amount of unknowns that depends on your antivirus, group policy settings any other programs on the same server daring to use docker as well and, from what I've seen, whether Jupiter is in Sagittarius.
PAW is significantly simpler and more straightforward on Linux, but you only get Centos or RHEL to choose from and older versions of both. And not every organisation has linux admins or allows linux deployments.

[7Zark7]

Thanks guys for your input - its much apprreciated and very helpful.

Looks like linux is the go under those particular versions mentioned below being Centos or RHEL. Based on what has been mentioned below in regards to instability in the windows server 2016 environment and the non support on 2019, windows sounds like an absolute nightmare. I guess the next hurdle is getting linux support from internal IT... should be an interesting conversation.

Cheers

Zark

[Alan Kirk]

Looks like linux is the go under those particular versions mentioned below being Centos or RHEL. Based on what has been mentioned below in regards to instability in the windows server 2016 environment and the non support on 2019, windows sounds like an absolute nightmare. I guess the next hurdle is getting linux support from internal IT... should be an interesting conversation.

Yes, it's one that I've started as well after this thread. I'm not sure how far I'll get, especially since it would require a non-current version.

Just remember, though, that the discussions above ONLY relate to running PAW. It's a bloody stupid situation, but there it is. You don't have to (and in my case I really don't want to) shift the TM1 server itself onto a Linux platform; to be clear, the version 11 (Planning Analytics 2) server itself runs just fine on Windows (anything from 2008 R2 on) and I don't want to shift everything that is Windows-based over to a Linux environment, not least because of case sensitivity issues. (Windows is not case sensitive, Linux is. Trust me when I tell you that this is a "can bite you in the @$$, badly" situation.) In another recent thread it was mentioned that the cloud version of PA2 uses the same arrangement; the TM1 server is on Windows, the PAW server is on Linux. (Which in itself speaks to how "well" the PAW solution was implemented for Windows.)

[keeponcubing]

While installing and managing PAW on Linux (CentOS7) is so much easier, faster, greater hair retention than Windows, we came across a random issue.

PH29529 UNABLE TO START WORKSPACE ADMIN TOOL WHEN /TMP IS MOUNTED WITH NOEXEC

I'm surprised since that IBM acknowledged the issue as it was disregarded for CA.

https://ibm-data-and-ai.ideas.aha.io/ideas/CAOP-I-2206

[jim wood]

...that is until we found out that you don't need a Workspace license to install it if you're only using it for PAx. After all certain CA licenses come with PAx licenses built in, so how else would you be able leverage it?

The other caveat is that Workspace license is only required for end users. As Workspace is the new admin / modelling environment to replace Architect (and Performance Mudler) any users who have an Admin license have automatic rights to use Workspace.

Good to know and thank you Lotsaram.

[Steve Rowe]

Good thread with many good points! (Especailly hierarchy aware formulas! After all these years....sigh)

We pretty much exclusively put clients onto a RHEL 7 build at this time, just to back-up the comments about the windows build.

The concept of PAW delivering the shared objects for Pafe is showing benefits at some of our customers.

The updates to the set editor and cube viewer are made to the PAW server, so with a single central update it is possible to get new / revised functionality to the client installs.

Where there is a "big IT" production made of updating any user side applications, with release packages, test cycles and client installs in their 100's, this delivers a clear benefit.

[Alan Kirk]

The concept of PAW delivering the shared objects for Pafe is showing benefits at some of our customers.

The updates to the set editor and cube viewer are made to the PAW server, so with a single central update it is possible to get new / revised functionality to the client installs.

Where there is a "big IT" production made of updating any user side applications, with release packages, test cycles and client installs in their 100's, this delivers a clear benefit.

That is indeed a very good point... but if, and only if, one accepts that it is either desirable or appropriate to design client software on the same basis as a piece of improv theatre and just make s**t up as you go along while hoping that it doesn't blow up.

In a notification sent out today, IBM stated:

Planning Analytics Workspace, Planning Analytics for Microsoft Excel and Planning Analytics Spreadsheet Services Release updates are released independently on a short cadence (SC) cycle of typically once per month. IBM Planning Analytics updates are released on a long cadence (LC) cycle, typically once per 3 months.

OK, disgusting, but OK, this isn't new. But we also have this:

As of the 2.0.55 SC release of IBM Planning Analytics, new versions of TM1 Web will be released on a more frequent schedule. New versions will be available approximately once a month, similar to the release schedules of IBM Planning Analytics Workspace and IBM Planning Analytics for Microsoft Excel.

TM1 Web is no longer included in the 2.0.x LC releases of Planning Analytics.

Call me old fashioned but I've always taken it as an article of faith that this is what software development should look like:
(a) You have an idea;
(b) You make a plan to develop all of the features that you want to incorporate;
(c) You write the code according to the plan;
(d) You test the code;
(e) You go through alpha and beta testing;
(f) You release the software;
(g) You fix the inevitable bugs in point releases;
(h) You listen to user feedback and incorporate new features in major releases.

In this way the user knows what to expect from the software at any given time.

We are not working on cutting edge particle physics at CERN here. When you strip away all of the mystical bulldust of BI ("Cloud!" "AI!", "X Analytics!" "Decision Intelligence!", "Predictive analytics!", better still "Prescriptive Analytics!"), PAFE, PAW and TM1 Web are nothing more than software that queries and/or updates a database, and there are only so many ways you can do that.

Once you pick one, there are only so many ways of writing the software and UI needed to do it.

And if you do it well, it's done, with only occasional tweaking required.

Which raises the question of WHY THE **** DO THEY EVEN NEED TO BE DOING MONTHLY RELEASES, which then need the "advantage" of having UI components picked up from a central deployment?

Are there so many bugs that it requires monthly updates?

Don't answer that.

Did IBM have a complete failure of imagination on step (b), and every couple of weeks someone says "OH wait, I know! Let's create a New! User! Experience!" (And to hell with all of the clients who aren't techies, and just want to get their work done, without needing to re-learn the UI for these "improvements".)

Did they have a complete failure on step (e)?

Again, don't answer that.

As much of a dinosaur as Perspectives is, how often did we need updates to it? (Other than the ones that we never got, but that's a different discussion.) This new and exciting ability to update UI components every month requires you to take it on faith that IBM haven't fracked something up in the latest "short cadence" update because, let's be brutally honest, such things are rarely seen until they go into mass usage which is why most companies don't (and shouldn't) go updating software every fracking month. It's a productivity tool, not a Wordpress site.

If a company builds the thing well in the first place, they don't NEED to be doing never ending release packages for IT to deploy.

Oh, and let's not forget... PAFE itself is included in this "short cadence" meaning that there is STILL a need to deploy new software on the client side, making the incorporation of UI components into PAW... not so much of an advantage.

[moby91]

If there is existing documentation found in the tips sections / useful IBM documentation and any caveats I should be aware of when upgrading I would be greatly appreciative. Any gotchas that I should be aware of i.e addtional ram consumption required etc from user experience would he helpful as well.

https://community.ibm.com/accelerators/ ... -Customers
IBM Accelerator Catalog
IBM TM1 to IBM Planning Analytics Local Upgrade Guide for IBM Customers
This document features guidelines for the upgrade from IBM Cognos TM1 to IBM Planning Analytics Local (on Premises). It introduces IBM Planning Analytics and differentiates IBM Planning Analytics from IBM Cognos TM1.


https://community.ibm.com/accelerators/ ... mendations
IBM Accelerator Catalog
Migration and Transitioning from TM1 to Planning Analytics - Guidance and Recommendations
This document describes how to migrate from TM1 10.2.2 to IBM Planning Analytics.


https://community.ibm.com/accelerators/ ... -Customers
IBM Accelerator Catalog
IBM TM1 On-Premises to IBM Planning Analytics Transitioning Guide for IBM Customers
This document features guidelines for the conversion and transition from IBM Cognos On-Premises or IBM Planning Analytics Local to IBM Planning Analytics Cloud.